With all the noise from the lead-up to the 2024 US Presidential election next week, you’d be forgiven for having missed some of the trailers that came out over the last few days. Not to worry; I’ve got you covered for some of the best ones.

Squid Game is finally returning for its second season soon, and in a new teaser, Lee Jung-jae’s Gi-hun looks like he’s preparing to stage an uprising. Whatever happens, it looks like the season will be a bloody, violent return for the dystopian competition on December 26th.

I’m very excited that Jude Law’s Star Wars debut, Star Wars: Skeleton Crew, is essentially a swashbuckling pirate adventure. The vibes it’s throwing off feel like a nice blend of The Goonies and Star Trek: Prodigy, and I’m certainly ready. The series debuts December 3rd on Disney Plus.

The moody trailer for Steven Soderbergh’s haunted house horror film Presence doesn’t show the supernatural figure that’s terrorizing Lucy Liu and her suburban family. That’s because it all takes place from the perspective of the ghost itself, which I didn’t realize until I read Variety’s write-up of this trailer. It’s already an eerie preview, but knowing that made it much more so on my second viewing. The movie hits theaters on January 24th.

Luca Guadagnino’s Queer, based on William Burroughs’ semi-autobiographical novel of the same name, stars Daniel Craig as William Lee, an American expat who lives in 1950s Mexico City and falls for a younger man named Eugene (Drew Starkey). Having already gotten showings at the Venice Film Festival, some reviews have praised it for Daniel Craig’s performance and Guadagnino’s “trippy” direction. Queer releases on November 27th.

From streamlining operations to automating complex processes, AI has revolutionized how organizations approach tasks – however, as the technology becomes more prevalent, organizations are discovering the rush to embrace AI may come with unintended consequences.

A report by Swimlane reveals while AI offers tremendous benefits, its adoption has outpaced many companies’ ability to safeguard sensitive data. As businesses deeply integrate AI into their operations, they must also contend with the associated risks, including data breaches, compliance lapses, and security protocol failures.

Agatha All Along is one of the most widely liked titles that Marvel Studios has released in, well, a while. The WandaVision spin-off premiered in late September and did a lot to win over even some of the Marvel Cinematic Universe’s more skeptical fans across its nine episodes. While Agatha All Along does set up some exciting future possibilities for several of its characters, though, its finale doesn’t include a single post-credits scene.

According to Agatha All Along creator Jac Schaeffer, that isn’t because she didn’t have any ideas for one. When asked about the series’ lack of a post-credits tag, Schaeffer told Variety, “That’s a Marvel decision. I know nothing more than that.” The writer and showrunner went on to reveal that she actually wrote multiple potential post-credits scenes for Agatha All Along, none of which were ultimately used because of behind-the-scenes decision-making by Marvel.

“I wrote a number of tags, because you always do on every Marvel everything. I love writing tags. I think some of my best writing is in the tags that were never made. I should have a little binder of my tags. They’re so fun to write, because you’re writing the promise without having to deliver on anything. They’re the best,” Schaeffer commented. “But there are so many things that factor into those. And I was told that we weren’t going to do a tag on this show.”

Schaeffer, of course, has experience writing post-credits scenes. After all, her previous Marvel show, 2021’s WandaVision, ends with a brief scene that directly sets up the events of 2022’s Doctor Strange in the Multiverse of Madness and specifically Wanda Maximoff’s (Elizabeth Olsen) villainous actions throughout it. The absence of a similar post-credits tag at the end of Agatha All Along, therefore, came as a shock for a number of reasons.

Marvel’s thinking behind this absence may have to remain a mystery for the time being to viewers. Fortunately, Agatha All Along does go out of its way to set up more adventures for, at the very least, Billy Maximoff (Joe Locke). The character teams up with the ghost of Agatha Harkness (Kathryn Hahn) to go searching for his missing brother’s soul in the Agatha All Along finale, but fans will have to wait to learn more about Marvel’s plans for Agatha, Billy, and the other members of the Maximoff family.

Agatha All Along is streaming now on Disney+.

Bose is without a doubt one of the top contenders for active noise-cancelling headphones with its QuietComfort model, and right now Amazon has a pretty good deal on them worth looking into.

Normally the QuietComfort headphones would retail for $349 at their full price. However, Amazon currently has them on sale for $199. This is a great price for these headphones and it’s the lowest we’ve seen them. It’s also the lowest price for these headphones in the last 30 days. With the average price sitting at around $285.20.

The main feature here is the active noise cancellation. It’s part of why these have been so popular over time. Because you can wear them and drown out almost everything so you can enjoy your music or whatever else you’re listening to. They’re great for travel in this sense. If you fly a lot, then these are perfect for taking on flights to make sure you don’t have to hear everyone else on the plane.

They also sound pretty good. Battery life is great too, lasting up to 24 hours on a single charge. While that isn’t the longest battery life we’ve ever seen, it’s more than enough for most people and will get you through a few days before you need to plug them in. A feature that we really love is that they fold up and they come with a protective travel case. Whether you use the case or not, they’re easily packable. Bose also sells these in several different colors. This includes Cypress Green, Moonstone Blue, Black, Blue Dusk, Chilled Lilac, Sandstone, and White.

Plus, all of these colors are on sale for the discounted $199 price tag. There’s an EQ in the companion app as well if you want to tune the sound to your personal liking.

Buy at Amazon

The FBI issued a statement on Saturday about deceptive videos circulating ahead of the election, saying it’s aware of two such videos “falsely claiming to be from the FBI relating to election security.” That includes one claiming the FBI had “apprehended three linked groups committing ballot fraud,” and one about Kamala Harris’ husband. Both depict false content, the FBI said.

Disinformation — including the spread of political deepfakes and other forms of misleading videos and imagery — has been a major concern in the leadup to the US presidential election. In its statement posted on X, the FBI added:

Just a day earlier, the along with the Office of the Director of National Intelligence (ODNI) and the Cybersecurity and Infrastructure Security Agency (CISA) said they’d traced two other videos back to “Russian influence actors,” including one “that falsely depicted individuals claiming to be from Haiti and voting illegally in multiple counties in Georgia.”

The introduction of ChatGPT has brought large language models (LLMs) into widespread use across both tech and non-tech industries. This popularity is primarily due to two factors:

1. LLMs as a knowledge storehouse: LLMs are trained on a vast amount of internet data and are updated at regular intervals (that is, GPT-3, GPT-3.5, GPT-4, GPT-4o, and others);

2.  Emergent abilities: As LLMs grow, they display abilities not found in smaller models.

Does this mean we have already reached human-level intelligence, which we call artificial general intelligence (AGI)? Gartner defines AGI as a form of AI that possesses the ability to understand, learn and apply knowledge across a wide range of tasks and domains. The road to AGI is long, with one key hurdle being the auto-regressive nature of LLM training that predicts words based on past sequences. As one of the pioneers in AI research, Yann LeCun points out that LLMs can drift away from accurate responses due to their auto-regressive nature. Consequently, LLMs have several limitations:

• Limited knowledge: While trained on vast data, LLMs lack up-to-date world knowledge.
• Limited reasoning: LLMs have limited reasoning capability. As Subbarao Kambhampati points out LLMs are good knowledge retrievers but not good reasoners.
• No Dynamicity: LLMs are static and unable to access real-time information.

To overcome LLM’s challenges, a more advanced approach is required. This is where agents become crucial.

Agents to the rescue

The concept of intelligent agent in AI has evolved over two decades, with implementations changing over time. Today, agents are discussed in the context of LLMs. Simply put, an agent is like a Swiss Army knife for LLM challenges: It can help us in reasoning, provide means to get up-to-date information from the Internet (solving dynamicity issues with LLM) and can achieve a task autonomously. With LLM as its backbone, an agent formally comprises tools, memory, reasoning (or planning) and action components.

Components of AI agents

• Tools enable agents to access external information — whether from the internet, databases, or APIs — allowing them to gather necessary data.
• Memory can be short or long-term. Agents use scratchpad memory to temporarily hold results from various sources, while chat history is an example of long-term memory.
• The Reasoner allows agents to think methodically, breaking complex tasks into manageable subtasks for effective processing.
• Actions: Agents perform actions based on their environment and reasoning, adapting and solving tasks iteratively through feedback. ReAct is one of the common methods for iteratively performing reasoning and action.

What are agents good at?

Agents excel at complex tasks, especially when in a role-playing mode, leveraging the enhanced performance of LLMs. For instance, when writing a blog, one agent may focus on research while another handles writing — each tackling a specific sub-goal. This multi-agent approach applies to numerous real-life problems.

Role-playing helps agents stay focused on specific tasks to achieve larger objectives, reducing hallucinations by clearly defining parts of a prompt — such as role, instruction and context. Since LLM performance depends on well-structured prompts, various frameworks formalize this process. One such framework, CrewAI, provides a structured approach to defining role-playing, as we’ll discuss next.

Multi agents vs single agent

Take the example of retrieval augmented generation (RAG) using a single agent. It’s an effective way to empower LLMs to handle domain-specific queries by leveraging information from indexed documents. However, single-agent RAG comes with its own limitations, such as retrieval performance or document ranking. Multi-agent RAG overcomes these limitations by employing specialized agents for document understanding, retrieval and ranking.

In a multi-agent scenario, agents collaborate in different ways, similar to distributed computing patterns: sequential, centralized, decentralized or shared message pools. Frameworks like CrewAI, Autogen, and langGraph+langChain enable complex problem-solving with multi-agent approaches. In this article, I have used CrewAI as the reference framework to explore autonomous workflow management.

Workflow management: A use case for multi-agent systems

Most industrial processes are about managing workflows, be it loan processing, marketing campaign management or even DevOps. Steps, either sequential or cyclic, are required to achieve a particular goal. In a traditional approach, each step (say, loan application verification) requires a human to perform the tedious and mundane task of manually processing each application and verifying them before moving to the next step.

Each step requires input from an expert in that area. In a multi-agent setup using CrewAI, each step is handled by a crew consisting of multiple agents. For instance, in loan application verification, one agent may verify the user’s identity through background checks on documents like a driving license, while another agent verifies the user’s financial details.

This raises the question: Can a single crew (with multiple agents in sequence or hierarchy) handle all loan processing steps? While possible, it complicates the crew, requiring extensive temporary memory and increasing the risk of goal deviation and hallucination. A more effective approach is to treat each loan processing step as a separate crew, viewing the entire workflow as a graph of crew nodes (using tools like langGraph) operating sequentially or cyclically.

Since LLMs are still in their early stages of intelligence, full workflow management cannot be entirely autonomous. Human-in-the-loop is needed at key stages for end-user verification. For instance, after the crew completes the loan application verification step, human oversight is necessary to validate the results. Over time, as confidence in AI grows, some steps may become fully autonomous. Currently, AI-based workflow management functions in an assistive role, streamlining tedious tasks and reducing overall processing time.

Production challenges

Bringing multi-agent solutions into production can present several challenges.

• Scale: As the number of agents grows, collaboration and management become challenging. Various frameworks offer scalable solutions — for example, Llamaindex takes event-driven workflow to manage multi-agents at scale.
• Latency: Agent performance often incurs latency as tasks are executed iteratively, requiring multiple LLM calls. Managed LLMs (like GPT-4o) are slow because of implicit guardrails and network delays. Self-hosted LLMs (with GPU control) come in handy in solving latency issues.
• Performance and hallucination issues: Due to the probabilistic nature of LLM, agent performance can vary with each execution. Techniques like output templating (for instance, JSON format) and providing ample examples in prompts can help reduce response variability. The problem of hallucination can be further reduced by training agents.

Final thoughts

As Andrew Ng points out, agents are the future of AI and will continue to evolve alongside LLMs. Multi-agent systems will advance in processing multi-modal data (text, images, video, audio) and tackling increasingly complex tasks. While AGI and fully autonomous systems are still on the horizon, multi-agents will bridge the current gap between LLMs and AGI.

Abhishek Gupta is a principal data scientist at Talentica Software.